Rock weathering creates oases of life in a High Arctic desert

Rock weathering creates oases of life in a High Arctic desert

Sara Borin,

Stefano Ventura,

Fulvia Tambone,

Francesca Mapelli,

Florence Schubotz,

Lorenzo Brusetti,

Barbara Scaglia,

Luigi P. D’Acqui,

Bjørn Solheim,

Silvia Turicchia,

Ramona Marasco,

Kai-Uwe Hinrichs,

Franco Baldi,

Fabrizio Adani

and Daniele Daffonchio

*

1

Department of Food Science and Microbiology, University of Milan, I-20133 Milan, Italy.

2

Istitute of Ecosystem Study, CNR, I-50019 Sesto Fiorentino, Italy.

3

Department of Crop Science, University of Milan, I-20133 Milan, Italy.

4

Department of Geosciences, University of Bremen, D-28359 Bremen, Germany.

5

Department of Biology, University of Tromsø, N-9037 Tromsø, Norway.

6

Department of Environmental Sciences, University of Venice ‘Ca’ Foscari’, I-30121 Venice, Italy.

Summary

During primary colonization of rock substrates by plants, mineral weathering is strongly accelerated under plant roots, but little is known on how it affects soil ecosystem development before plant establish- ment. Here we show that rock mineral weathering mediated by chemolithoautotrophic bacteria is asso- ciated to plant community formation in sites recently released by permanent glacier ice cover in the Midtre Lovénbreen glacier moraine (78°53 ⬘N), Svalbard.

Increased soil fertility fosters growth of prokaryotes and plants at the boundary between sites of intense bacterial mediated chemolithotrophic iron-sulfur oxi- dation and pH decrease, and the common moraine substrate where carbon and nitrogen are fixed by cyanobacteria. Microbial iron oxidizing activity deter- mines acidity and corresponding fertility gradients, where water retention, cation exchange capacity and nutrient availability are increased. This fertilization is enabled by abundant mineral nutrients and reduced forms of iron and sulfur in pyrite minerals within a

conglomerate type of moraine rock. Such an interac- tion between microorganisms and moraine minerals determines a peculiar, not yet described model for soil genesis and plant ecosystem formation with potential past and present analogues in other harsh environments with similar geochemical settings.

Introduction

The factors affecting initial steps of soil formation, ecosys- tem development and plant establishment in extreme arid environments like cold and hot deserts, where the avail- ability of water and nutrients limits biomass growth (Sundareshwar et al., 2003; Yoshitake et al., 2007), are poorly constrained. In arid systems, organismic re- sponses result from overcoming a threshold in water availability, which is key limiting factor (Ogle and Rey- nolds, 2004; Schwinning et al., 2004; López et al., 2008).

Such responses may trigger dramatic changes and, after a cascade of events, finally manifest in comparably immense increases of biomass in soil in conjunction with increase in biological diversity and profound changes in the landscape. Such a behaviour holds also true for microbial life, which include true primary colonizers either autotrophic like photosynthetic cyanobacteria, or het- erotrophic microorganisms (Hodkinson et al., 2002). For instance in biological soil crusts, certain cyanobacteria quickly respond to water inputs and move actively towards the moist soil surface or refuge in the deeper soil layers when soil undergoes drying conditions (Garcia- Pichel and Pringault, 2005).

Mineral weathering in soil enhances availability of water and nutrients, leading to an increase of fertility (Anderson et al., 2000; Bashan et al., 2002; Puente et al., 2004a,b;

Konhauser, 2007; Yoshitake et al., 2007). Weathering occurs in hot and cold deserts (Ascaso et al., 1990;

Adams et al., 1992). During primary colonization of rock substrates, mineral weathering is strongly accelerated under plant roots (Bashan et al., 2002; Uroz et al., 2007) and enhanced by the association with rhizoplane micro- biota (Carrillo-Garcia et al., 1999; Bashan et al., 2000), but little is known on how weathering affects the soil ecosystem development before plant establishment.

Arctic glacier moraines recently released from perma- nent ice cover are ideal models for the assessment of factors associated with the initial development of a mature

correspondence. E-mail daniele.daffonchio@unimi.it; Tel. (+39) 250319117; Fax (+39) 250319238. ^†Present address: Faculty of Science and Technology, Free University of Bozen/Bolzano, Via Sernesi 1, 39100 Bolzano, Italy.

soil ecosystem. Within the grey-coloured moraine of the Midtre Lovénbreen glacier (78°53 ′N), Svalbard, between the Hodkinson proglacial chronosequence sites ML2 and ML3, which were released from the permanent ice cover 23 and 44 years ago respectively (Hodkinson et al., 2003), we found several sites densely colonized by the mosses Ditricum flexicaule and Bryum sp. and the vascu- lar plant Saxifraga oppositifolia (Hodkinson et al., 2003) (Fig. 1A and B). Green strips of flourishing vegetation were adjacent to rusty iron oxide-leaching strips sloping along the moraine with the latter departing from yellowish- rusty-greenish stones, rock conglomerates rich in reduced iron, in the form of pyrite. In a defined area of the moraine we found several of these rusty and green strips that were exclusively associated with one particular stone type.

Here, we present evidence that under harsh conditions of recently deglaciated moraines of the High Arctic, enhanced rock weathering mediated by chemolithoau- totrophic bacteria is associated to soil formation and increases of soil fertility that fosters the early establish- ment and growth of plant biocoenosis, the ecological com- munity of plants that sustains the ecosystem.

Results and discussion

We initially focused on one of these sites, named ML-RS1, estimated to be released by ice about 27 years ago. We observed several sites like ML-RS1 in the south- east portion of the moraine. We counted at least four sites similar to ML-RS1 with stones of similar size. In an area of about 1 hectar of the moraine we observed tens of smaller

A. A rusty leaching strip departs from a stone within the grey-coloured moraine (white bar= 50 cm). White arrows indicate zones where dense plant biomass can be observed. A detail of an area densely colonized by plants within the site is shown in (B) (white bar= 50 cm).

C. Biological markers measured along transect A, located 5 cm underneath the stone. C1, Absolute abundances of total IPLs) and sterols along transect A and the relative contribution of phosphatidylethanolamine (PE), phosphatidyl-(N)-methylethanolamine (PME) and phosphatidyl-(N,N)-dimethylethanolamine to total IPLs. C2, percentage of nitrogen and pH values in the soil crust.

D. Map of the area showing the position of the sampling transects.

sites with much smaller stones where signs of the plant flourishing like those in site ML-RS1 were present. We also found a larger site named ML-RC1 with the same but more extended phenomenon in place (Fig. S1). Plant flourishing occurred where the type of conglomerate rock like that of site ML-RS1 emerged to open air from the subsurface. For instance, site ML-RC1 looked like a large conglomerate vein cropping out from the grey moraine.

X-ray diffractometry and scanning electron microscopy- energy dispersive X-ray analyses of thin sections of the stone at site ML-RS1 showed pyrite microcrystals (Fig. 2).

The stone was a conglomerate consisting mainly of quartz, cemented by kaolinite, with additional contribu- tions of pyrite, aluminium-silicates, calcium hydroxyapa- tite and manganese oxides. On the surface, besides jarosite and gypsum, iron oxy-hydroxides and oxides could be detected as well as small amounts of organic carbon (0.3% w/w), supporting the presence of microbial biomass.

A higher density of plants was apparent in the green strip of site ML-RS1 compared with the surrounding greyish moraine. Plant and/or eukaryotic algal sterols, dominated by campesterol and sitosterol, were 10–100 times more concentrated in the green strips than in the surrounding moraine. The sterol concentrations declined to values close to zero in the rusty strip, suggestive of hostile conditions for plants and eukaryotic microorgan- isms (Fig. 1C1). Intact polar lipids (IPLs), representative of viable cells (Sturt et al., 2004; Biddle et al., 2006; Lipp et al., 2008), showed a similar pattern as the sterols, with highest concentrations in the green strips and lower values in the rusty and moraine areas. The sharp decline of sterols in the rusty area suggests that the IPLs in this zone largely represent bacterial biomass.

The composition of IPLs in the transect was diverse and represented a mixture of eukaryotic and bacterial biomass, largely derived from phototrophic organisms.

Abundant IPLs were betaine lipids linked to mono- and polyunsaturated fatty acids, glycoglycerolipids esterified to polyunsaturated fatty acids and phosphoglycerolipids with saturated and monounsaturated fatty acids (see Supporting information, Table S1, Fig. S4). This com- bination of lipids is typical of lower plants and arctic mosses (Dembitsky and R ˇ ezanka, 1995), cyanobacteria (R ˇ ezanka et al., 2003) and bacteria involved in plant symbiosis (López-Lara et al., 2003). Heterocyst glycolip- ids from nitrogen-fixing cyanobacteria (Nichols and Wood, 1968) were detected in high amounts throughout the transect (~20% of total IPLs), indicating that heterocyst cyanobacteria are evenly distributed in the green and rusty area (see Supporting information, Fig. S4). Towards the rusty strip, the phospholipids phosphocholine (PC), phosphoethanolamine (PE),

phospho-(N)-methylethanolamine (PME) and phospho- (N,N)-dimethylethanolamine (PDME) increase (Fig. 1C1).

These lipids are commonly observed in cultures of acido- philic bacteria, e.g. Acidithiobacillus ferrooxidans and Acidiphilium (Barridge and Shively, 1968; Harms, 2007), and thus are consistent with an increase of these bacteria in the rusty strip. Ornithine lipids and bacteriohopanepoly- ols bacteriohopane-32,33,34,35-tetrol and 35-amino- bacteriohopane-32,33,34-triol (aminotriol), which are exclusive bacterial markers (Lopez-Lara et al., 2003;

Talbot et al., 2008), were also detected in low abundances throughout the transect, however, with highest amounts outside the rusty area. We assign the presence of these lipids to bacteria associated with plant-symbiosis of the order Rhizobiales, which were detected in the green area (Fig. 3, Table S4). In summary, the elevated biomass content adjacent to the rusty strip underscores the link between increased fertility by rock weathering and the establishment of an oasis of life.

To examine microenvironmental conditions promoting plant growth, we established a grid pattern for a detailed sampling of site ML-RS1 (Fig. 1D) along mini-transects at increasing distance from the stone. Total organic carbon, nitrogen percentage, potential nitrogen fixation and respi- ratory activity were higher in the green than in the rusty strip, but similar to those in the surrounding greyish moraine (Fig. 1C2, Table S2). Stable isotopic composi- tions of nitrogen d

N, of slightly below 0‰ indicated that in all transects bacterial nitrogen fixation is the main source of nitrogen (Högberg, 1997). The presence of het- erocyst glycolipids in all samples (Fig. S4) strongly sup- ports this interpretation. We conclude that nitrogen supply is provided to both the green and the greyish soil crusts and is probably not limiting plant growth in the green strip.

We determined parameters indicative of soil fertility in the grid pattern of site ML-RS1 after removal of the plant cover (Table S2). Mean values of fine particles, cation exchange capacity (CEC), water holding capacity (WHC), exchangeable Mg, K and Fe and Fe

were significantly higher in the green than in the greyish crusts. Relative increments, for the whole soil crust (Table 1), were between 36.2% (K

) and 132% (Fe

), with CEC and WHC, respectively, increasing of 95.2% and 48.1%. Nutri- ent differences are particularly meaningful considering that in the green strip plants act as element scavengers and bioaccumulators that continuously extract nutrients from the soil crust. The increased WHC markedly sup- ports more favourable conditions for plant growth in this desert ecosystem that receives an average precipitation of 190 mm year

(range, 88–265 mm in the period 1988–

2002; http://www.svalbard.com/weather.html). Such low

values indicate that water is the primary limiting factor for

development of plant biocoenoses (Ogle and Reynolds,

2004; Lopez et al., 2008).

Fig. 2. Pyrite microcrystals and oxidation signatures on the surface of the stone upstream the rusty strip of site ML-RS1.

A and C. SEM-EDX back scattering electron micrographs of thin-layer shiny sections of the stone, showing pyrite and apatite microcrystals (circled) respectively. Peaks attributable to the different elements of pyrite and apatite crystals, circled in (A) and (C), are shown in (B) and (D) respectively.

E. X-ray diffractograms of the surface and the internal part of the conglomerate. Diffractograms of the yellowish and rusty coatings and the internal part of the stone are shown in red, green and blue colours respectively. Marked peaks are G, gypsum; J, jarosite; K, kaolinite; P, pyrite; Q, quartz.

A low pH was measured (Fig. 1C, Table S2) in the rusty strip (down to 3.5; mean value along the rusty vertical transect, 5.2 ⫾ 1.4, n = 10) and presented a raising gradient in the green strips (7.1 ⫾ 0.3, n = 25) to reach neutral-subalkaline values in the greyish moraine (7.6 ⫾ 0.4, n = 9). pH is significantly correlated with both

total (Fe

) and exchangeable (Fe

) iron contents (r = -0.959, P < 0.01, n = 15, and r = -0.932, P < 0.01, n = 15 respectively). We hypothesized that ferrous iron, present in the form of pyrite, was dissolved and trans- ported from the conglomerate stone, and subsequently oxidized to form the acidic rusty strip. Fe

oxidation was

Each graph shows the relative composition of the bacterial community inhabiting the soil crust. A4 and B4 were samples from the rusty soil crust, A6 and B6 from the green soil crusts and A9 and B9 from the greyish soil crusts. Ninety clones have been randomly sequenced from each library. Sequencing primer started from position 27 and the sequence length was 700–900 bp.

Table 1. Parameters indicative of soil fertility in the greyish and green soil crusts of site ML-RS1 and relative percentage increments in the green respect to the greyish crusts over the whole soil crust (fine plus coarse particle fractions).

Parameters

Soil crust type^a

Relative increment^b (%)

Greyish Green

CEC (cmol⁽⁺⁾kg^-1dm) 2.9⫾ 0.9 5.7⫾ 1.5 95.2

WHC (g 100 g^-1dm) 21.6⫾ 2.9 32.0⫾ 8.7 48.1

Feexch(mg kg^-1dm) 65.4⫾ 27.8 151.8⫾ 28.5 132.0

Fe²⁺(mg kg^-1dm) 34.2⫾ 6.2 50.0⫾ 5.4 46.4

Mgexch(mg kg^-1dm) 73.4⫾ 20.4 113.6⫾ 15.8 54.7

Kexch(mg kg^-1dm) 103.2⫾ 29.0 140.8⫾ 35.1 36.2

a. All means are significantly different (P< 0.05) according to Duncan test.

b. Relative percentage increments of parameters in the green crust respect to the greyish crust.

exch, exchangeable.

corroborated by the lower Fe

content in the rusty than in the green strips, despite Fe

and Fe

contents were significantly higher (Table S2). The hypothesis is consis- tent with the presence of jarosite [KFe

(SO

)

(OH)

] on the stone surface and within the rusty strip as shown by Diffuse Reflectance Infrared Fourier Transformed (DRIFT) spectroscopy (Fig. S2A). Jarosite has been used as a water signature on Mars (Klingelhofer et al., 2004), but is unstable and rare under oxic conditions on Earth (Darmody et al., 2007). Jarosite can be formed by spon- taneous or microbially mediated oxidation of pyrite (Sasaki and Konno, 2000). The evidence at site ML-RS1 is consistent with such a scenario.

The observed physicochemical differences between the rusty, green and greyish strips are reflected in the structures of the respective bacterial communities. Bac- terial diversity was higher in the green and the greyish than in the rusty strips that appeared as the least equi- table as shown by Shannon–Wiener and Equitability indices of bacterial amplified ribosomal intergenic spacer analysis (ARISA) (Table S3). Detrended correspondence analysis of ARISA profiles tended to cluster the samples according to the soil crust pH, indicating that different bacterial communities were present in the different soil crusts (Fig. S3). Soil crust pH was strongly correlated with bacterial diversity (ARISA detrended correspon- dence analysis axis 1 versus pH, r = 0.835, P < 0.001, Fig. S3). Soil pH has already been shown as a major driver of biogeographical repartition of bacteria in soil (Fierer and Jackson, 2006). To identify bacteria inhabit- ing the soil crusts, clone libraries of amplified 16S rRNA genes were generated from six soil crust samples within the grid pattern, two for each colour (rusty, greenish and greyish strips) in two horizontal transects sampled at 5 and 23 cm from the stone originating the leaching strips. A total of 540 clones were sequenced (Fig. 3, Table S4). The rusty soil crusts were dominated by a-Proteobacteria of the genus Acidiphilium (order Rho- dospirillales; 30–32% of the clones in the two libraries respectively) and by Acidobacteria (37–39% of the clones respectively), both frequently associated to acidic environments where oxidation of iron and sulfur occur (Gonzàlez-Toril et al., 2003; Diaby et al., 2007). The greyish soil crusts were dominated by Cyanobacteria (29–40% of the clones respectively) followed by a-Proteobacteria mostly affiliated with orders other than Rhodospirillales. In contrast, green soil crusts showed a bacterial community typical of plant-inhabited soils.

Hence, at the boundary between a cyanobacterial carpet typical of substrates recently released by permanent ice cover (Nemergut et al., 2007), and iron weathering spots, the geochemical conditions selected a rhizosphere-like bacterial community able to support a plant biocoenosis. We suggest that the major driver of

such community transition was the pH gradient resulting from the activity of iron-oxidizing bacteria.

By inoculating an organic carbon-free medium added with reduced iron sulfate either with the stone surface or the rusty soil crust, we isolated rod-shaped bacteria able to grow at 4°C and pH 2 and to produce jarosite (Fig. S2B). The isolates were identified as A. ferrooxidans by 16S rRNA gene sequencing and contained phosphatidyl-(N)-methylethanolamine (94% of total IPLs) as a major lipid type, next to phosphatidylglycerol and ornithine lipids (Fig. S4), all of which are present in the rusty strip and surrounding area. At a second site (ML- RC1) that resembles site ML-RS1 (Fig. S1), A. ferrooxi- dans counts decreased from 1.7 ¥ 10

cells g

close to the rock to 3.0 ¥ 10° at about 10 m distance from the leaching origin. Using the major lipid PME of A. ferrooxi- dans as a proxy for estimating this organism’s contribution to the microbial population, we infer that in ML-RS1 site it accounts not more than 0.8% of the total community in transects A and B (Fig. S4). The mean abundance of the PME derivatives found in A. ferrooxidans decreases with distance from the rock, from ~0.4% (transect A) to ~0.2%

(transect B) as cell counts decrease. This is consistent with the failure of 16S rRNA gene libraries in detecting this species. Acidithiobacillus ferrooxidans gains energy upon iron oxidation, one of the less energy-releasing metabolic processes (DG°

= -27.2 KJ mol

, pH 2.5 Konhauser, 2007). Taking into account different estimates of Acidithio- bacillus efficiency in the reaction, it was calculated that it would need to oxidize between 50 and 500 mol of ferrous iron to assimilate one mol of carbon (Ehrlich and Newman, 2008). As a consequence, A. ferrooxidans cell abundance can be very low, despite high iron oxidation activity.

In summary, specific sites within the Midtre Lovénbreen

moraine, where stones rich in reduced iron and sulfur are

present, provide conditions suitable to support growth and

activity of an acidophilic iron-oxidizing bacterial commu-

nity. Progressive stone fracturing and surface desegrega-

tion due to winter freezing, and chemical and biological

weathering, and the water-mediated particle dragging

maintained a constant pyrite supply over time on the

substrate slope downstream the stone, allowing the

establishment of a chemolithoautotrophic bacterial com-

munity (Fig. 4A). Stone leaching during summer snow

melting created a gradient along the slope with decreas-

ing iron concentration and acidification, as shown by the

increasing pH along the rusty strip (Table S2). Oxidation

of reduced iron released from the stone caused jarosite

and ferric oxy-hydroxide formation responsible for the

increase of both soil crust specific surface area (SSA) and

CEC, the latter determining a higher WHC and nutrient

content in the green strip (Table 1; Fig. 4B). The positive

and negative correlations between SSA and total and

reduced iron contents, respectively, r = 0.835, P < 0.0001 (n = 15) and r = -0.599, P < 0.01 (n = 15), supported such a mechanism. Hence, with respect to the rest of the moraine, plant establishment was favoured by the combi- nation of increased soil crust fertility (Fig. 4C) driven by the acidification and leaching activity of a chemolitotrophic community strongly influenced by iron oxydizers, and by cyanobacteria-mediated primary productivity.

The synergy between these two bacterial autotrophic processes in favouring microorganism and plant coloniza- tion is probably not restricted to the Midtre Lovénbreen glacier foreland, but should be explored in other environ- ments where similar geochemical conditions exist (Ander- son et al., 2000; Skidmore et al., 2005; Darmody et al., 2007), such as natural pyritic belts (Fernàndez-Remolar et al., 2005) and manmade pyritic mine tailings (Southam and Beveridge, 1992). Such a synergy could also have played a role in the primordial land colonization by plants (Rensing et al., 2008) or in soil formation on extraterres- trial bodies where signatures of acidic aqueous systems with iron- and sulfur-based redox cycles have been found (Bibring et al., 2007).

Experimental procedures

Soil crusts in the moraine of the Midtre Lovénbreen glacier, Ny Ålesund, Svalbard, were collected during three expedi- tions in August 2004 and 2005 and early September 2006 within the sampling grid designed on site ML-RS1 (Midtre Lovénbreen Rusty Site 1; 78°53.753′N, 12°05.115′E; 57.1 m a.s.l.) and along the slope of site ML-RC1. The top 5 mm of the crusts were separated from the underlying mineral matter and the two fractions were separately processed. Recovered samples were stored at-20°C for chemical and nucleic acid- based studies, or at 4°C for microbiological enrichments or activity measurements.

Soil crust and stone chemical analysis

Soil parameters were determined as follows: soil crusts pH in aqueous solution using a 1:2.5 sample/water ratio; fine (< 2 mm) and coarse (> 2 mm) particles by soil sieving after drying; total nitrogen by Kjeldahl method; available-P by using the Olsen and Bray and Kurtz methods depending on the pH; organic carbon by wet oxidation. For CEC and exchangeable Ca, Mg, K, Na determinations, samples were saturated with BaCl2-Triethanolamine solution (pH 8.1) and exchangeable cations were determined by Inductively Fig. 4. A schematic of the overall process occurring in ML-RC1 site in the moraine of the Midtre Lovénbreen glacier, Svalbard.

Coupled Plasma (ICP-MAS VARIAN, Liberty AX, Walnut Creek, CA). Total Ca, Mg, K, Na, Fe and Al were determined by samples digestion with HNO3(16 mol l^-1) in a microwave furnace (CEM Mars 5, Matthevs, North Caroline), and suc- cessive detection by Inductively Coupled Plasma. All the above methods were as described in Pansu and Gautheyrou (2006). WHC was determined by the Stackman box method (Klute, 1986). Exchangeable iron forms were extracted with EDTA or DTPA depending on crust pH (Lindsay and Norwell, 1969; Lakanen and Erviö, 1971) and determined with an Inductively Coupled Plasma. Reduced iron forms were extracted with the coloured EDTA-BPDS solution and quan- tified spectrophotometrically at 535 nm (Wang and Peverly, 1998). SSA of soil crust samples were determined by N2

adsorption (BET method) of dried samples by using a sorp- tometer apparatus (Quantachrome NovaWin2, NOVA Instru- ments, Boynton Beach, FL, USA). All analytical data were determined on the fine particle fraction (< 2 mm) (Pansu and Gautheyrou, 2006). Parameters with statistically significant differences between the greyish and the green crusts (Table S2) were also referred to the whole sample (Table 1) by taking into account the fine and the coarse particle con- tents of the whole soil crust and their standard deviations recalculated according to statistical standards (Ciani et al., 1990). All the analyses were made in triplicate and the results were analysed by ANOVA, considering the colour strips as independent variables. Values of the means were separated by Duncan test, considering a significance level of P< 0.05.

All statistical analyses were performed using theSPSS13.0 software package (SPSS International, Chicago, IL, USA).

Stable nitrogen isotopes were determined from 20 to 30 mg of homogenized soil on an elemental analyser (Ther- moQuest EA/NA 1110) coupled to a Thermo Finnigan Delta Plus isotope ratio mass spectrometer via a ConFlow II inter- face. Analyses were calibrated with a known standard (Wadden Sea soil) and were normalized to the international standard IAEA-CH-6, sucrose.

Shine thin sections of stone samples for Environmental Scanning Electron Microscopy-Energy Dispersive X-ray spectroscopy were visualized by a scanning electron micro- scope (FEI Quanta mod. 200, Endhoven, Nl) at an accelera- tion voltage of 25 kV to determine elements in soil particles (triplicate) at 1.0 and 0.1 bar after 100 s scanning. Mean concentrations and standard deviations of each element were calculated from three random determinations at differ- ent spots on the samples. The X-ray beam was 4mm wide and penetrated to a depth of 2mm.

X-ray diffractometry was performed on samples collected from the surface of the conglomerate on the basis of the different coating colours, reddish and yellowish, and from the internal part of the rock. Due to the rock nature, it was easy to break and separate the first 2–3 mm of the rock. Powdered samples were analysed by a Philips PW 3830 X-ray diffrac- tometer using CoKa radiation incident from 3 to 75°q with a step size of 0.02°q and a scan step time of 1 s. Total C and N content were determined using a Carlo Erba (Milan, Italy) NA 1500 CHNS Analyser. To distinguish organic from inorganic C was used a procedure reported in Santi and colleagues (2006). Two aliquots of each sample, in two replicates, were analysed by dry combustion: the first without treatment, to assess the total C and N content and the other after treatment

with excess HCl, for carbonate removal, to assess only the organic C content.

Rock samples and jarosite were analysed by DRIFT spec- troscopy using an Avatar 370 FT-IR from ThermoNicolet Instruments (Madison, WI, USA). Samples previously dried at 65°C for 48 h, and KBr (FT grade, Aldrich Chemical Co, ST Louis, Missouri) were mixed in the 1:10 ratio (w/w) and finely ground for 10 min using an agate ball mill (Specamill- Greseby-Specac, Kent, UK). Instrument parameters used were: scanning 128, resolution 4 cm^-1, and frequency 400–

4000 cm^-1gain 16. Peaks assignments were made according to Sasaki and Konno (2000).

Lipid biomarkers

Lipid analysis followed the procedures outlined by Sturt and colleagues (2004). In brief, 5–10 g of freeze-dried soil was extracted using a modified Bligh and Dyer protocol. Total lipid extracts were analysed on a ThermoFinnigan LCQ Deca XP Plus HPLC-MS system in positive and negative ionization mode. Quantification of IPLs was achieved using an internal standard (1-O-hexadecyl-2-acetoyl-sn-glycero-3- phosphocholine) and external calibration of this standard. For the analysis of bacteriohopanepolyols, an aliquot of the total lipid extract was acetylated with acetic anhydride/pyridine (1:1) for 1 h at 50°C and subsequently measured on the same HPLC system using conditions described by Talbot and colleagues (2008). Sterols were analysed in an alcohol frac- tion that was isolated by solid-phase extraction and deriva- tized according to Hinrichs and colleagues (2000). The fractions were measured on a ThermoFinnigan Trace GC coupled to a Finnigan Trace MSplus and to a Finnigan DSQ.

Microbial community studies

Total DNA was extracted with the FastDNA SPIN Kit for Soil (BIO 101 Systems Q-BIO gene; CA, USA) following manu- facturer’s instructions. The ARISA fingerprinting was deter- mined as described elsewhere (Cardinale et al., 2004) and downstream multivariate statistical analyses were done with MVSP software (Kovach Computing Services, UK). 16S rRNA gene clone libraries from selected soil crust samples of site ML-RS1 were prepared as previously described (Van der Wielen et al., 2005). Ninety clones have been randomly chosen from each library. 700- to 900-bp-long sequences, corresponding to the first half of the 16S rRNA gene, were aligned with ClustalX (Thompson et al., 1997), and opera- tional taxonomic unit distribution was calculated with Vector NTI Advance 10 using a 97% identity threshold. Good cov- erage of the dominant operational taxonomic units within libraries was evaluated by rarefaction analysis with PAST software (Hammer et al., 2001).

Isolation and cultivation of chemolithoautotrophs

One gram of rusty soil crusts or stone surface was inoculated in carbon-free 9K medium supplemented with ferrous sulfate (Lizama and Suzuki, 1988) and incubated at 4°C for 4 weeks until a rusty precipitate of iron oxides was visible, and

bacterial cells were observable. Three subsequent steps of terminal dilution to extinction enabled obtaining pure cultures of rod-shaped bacteria. A 16S rRNA gene library from the culture was established, and the sequencing of several clones revealed identical sequences having 99% identity with the 16S rRNA gene of A. ferrooxidans. The isolated strains were cultured in modified 9K (Silverman and Lundgren, 1959) supplemented with either ferrous sulfate or powdered pyrite (0.5% W/V). At 28°C the strains produced jarosite after 3 days of incubation. Most probable number counts of autotrophic iron oxidizing bacteria were performed on soil crusts from site ML-RC1. Soil crusts were inoculated (1%

W/V) in decimal serial dilutions in modified 9K medium supplemented with iron (II) sulfate in 3 ml microtiter plates.

After 4 weeks of incubation at 15°C growth coincided with the appearance of a rusty precipitate in the wells.

Activity measurements

Potential nitrogen fixation activity of soil samples was mea- sured as described elsewhere (Zielke et al., 2002). Freshly collected 1-cm-thick top layer soil crust samples (surface area, 64 cm²) were conditioned at room temperature (24°C) in natural light and 0 water potential (water saturated) for 17 h prior to incubation with 10% acetylene (v/v) for 3 h in the same conditions.

Soil crust respirations were measured by trapping with alkali the CO2produced during soil crust incubation at 20°C in the laboratory for 21 days (Bekku et al., 1997). All the analy- ses were performed in triplicate.

Acknowledgements

The Authors thank Agostino Rizzi for performing SEM-EDX analysis of thin sections of the stone, Manuela Spagnol for surface area of soil crust samples, Valentina Orzi for respi- ration activity measurements, Silvia Salati for cation content determinations and Alessandro Dodero for CHNS analysis.

Financial support comes from the project FIRST 2007 ‘Studio della successione microbica lungo un transetto di deglaciazi- one del ghiacciaio Midtre Lovèn, Isole Svalbard (Norvegia)’

granted by the University of Milan. S.V. and S.T. acknowledge a travel grant from the ESF Scientific Programme on Nitrogen Fixing Cyanobacteria – CYANOFIX for a 1 month expedition to Svalbard. S.V. acknowledges a short-term mobility grant of CNR to travel to Svalbard. F.S. and K.U.H. acknowledge support by DFG (trough MARUM Center for Marine Environ- mental Sciences) and thank Monika Segl for help during nitrogen isotope analysis. Authors thank CNR-Polarnet for the use of the Italian Polar Station in Ny-Ålesund.

References

Adams, J.B., Palmer, F., and Staley, J.T. (1992) Rock weath- ering in deserts: mobilization and concentration of ferric iron by microorganisms. Geomicrobiol J 10: 99–114.

Anderson, S.A., Driver, J.I., Frost, C.D., and Holden, P.

(2000) Chemical weathering in the foreland of a retreating glacier. Geochim Cosmochim Acta 64: 1173–1189.

Ascaso, C., Sancho, L.G., and Rodriguez Pascual, C. (1990)

The weathering action of saxicolous lichens in maritime Antarctica. Polar Biol 11: 33–40.

Barridge, J.K., and Shively, J.M. (1968) Phospholipids of the Thiobacilli. J Bacteriol 95: 2182–2185.

Bashan, Y., Davis, E.A., Carrillo-Garcia, A., and Linderman, R.G. (2000) Assessment of VA mycorrhizal inoculum potential in relation to the establishment of cactus seed- lings under mesquite nurse-trees in the Sonoran Desert.

Appl Soil Ecol 14: 165–175.

Bashan, Y., Li, C.Y., Lebsky, V.K., Moreno, M., and de-Bashan, L.E. (2002) Primary colonization of volcanic rocks by plants in arid Baja California, Mexico. Plant Biol 4:

392–402.

Bekku, Y., Koizumi, H., Oikawa, T., and Iwaki, H. (1997) Examination of four methods for measuring soil respiration.

Appl Soil Ecol 5: 247–254.

Bibring, J.-P., Arvidson, R.E., Gendrin, A., Gondet, B., Lan- gevin, Y., Le Mouelic, S., et al. (2007) Coupled ferric oxides and sulfates on the martian surface. Science 317:

1206–1210.

Biddle, J.F., Fitz-Gibbon, S., Schuster, S.C., Brenchley, J.E., and House, C.H. (2006) Heterotrophic Archaea dominate sedimentary subsurface ecosystems off Peru. Proc Natl Acad Sci USA 103: 3846–3851.

Cardinale, M., Brusetti, L., Quatrini, P., Borin, S., Puglia, A.M., Rizzi, A., et al. (2004) Comparison of different primer sets for use in automated ribosomal intergenic spacer analysis of complex bacterial communities. Appl Environ Microbiol 70: 6147–6156.

Carrillo-Garcia, Á., León de la Luz, J.L., Bashan, Y., and Bethlenfalvay, G.J. (1999) Nurse plants, mycorrhizae, and plant establishment in a disturbed area of the Sonoran Desert. Restor Ecol 7: 321–335.

Ciani, G., Fusi, A., Nicotra, F., and Rindone, B. (1990) Labo- ratorio di Chimica. Napoli, Italy: Società Editrice Scientifica.

Darmody, R.G., Thorn, C.E., and Dixon, J.C. (2007) Pyrite- enhanced chemical weathering in Kärkevagge, Swedish Lapland. Geol Soc Am Bull 119: 1477–1485.

Dembitsky, V.M., and Rˇ ezanka, T. (1995) Distribution of dia- cylglycerylhomoserines, phospholipids and fatty acids in thirteen moss species from southwestern Siberia. Biochem Syst Ecol 23: 71–78.

Diaby, N., Dold, B., Pfeifer, H.R., Holliger, C., Johnson, D.B., and Hallberg, K.B. (2007) Microbial communities in a por- phyry copper tailings impoundment and their impact on the geochemical dynamics of the mine waste. Environ Micro- biol 9: 298–307.

Ehrlich, H.L., and Newman, D.K. (2008). Geomicrobiology.

Florida, USA: CRC Press.

Fernàndez-Remolar, D.C., Morris, R.V., Gruener, J.E., Amils, R., and Knoll, A.H. (2005) The Río Tinto Basin, Spain:

mineralogy, sedimentary geobiology, and implications for interpretation of outcrop rocks at Meridiani Planum, Mars.

Earth Planet Sci Lett 240: 149–167.

Fierer, N., and Jackson, R.B. (2006) The diversity and bio- geography of soil bacterial communities. Proc Natl Acad Sci USA 103: 626–631.

Garcia-Pichel, F., and Pringault, O. (2005) Cyanobacteria track water in desert soils. Nature 413: 380–381.

Gonzàlez-Toril, E., Llobet-Brossa, E., Casamaior, E.O., Amann, R., and Amils, R. (2003) Microbial ecology of an

extreme acidic environment, the Tinto river. Appl Environ Microbiol 69: 4853–4865.

Hammer, Ø., Harper, D.A.T., and Ryan, P.D. (2001) PAST:

paleontological statistics software package for education and data analysis. Palaeontol Electron 4: issue 1, art. 4.

Harms (2007) Diversity of intact polar lipids in an extreme acidic environment: the Rio Tinto, Spain. Diploma Thesis.

University of Bremen, Bremen, Germany.

Hinrichs, K.-H., Summons, R.E., Orphan, V., Sylva, S.P., and Hayes, J.M. (2000) Molecular and isotopic analyses of anaerobic methane-oxydizing communities in marine sedi- ments. Org Geochem 31: 1685–1701.

Hodkinson, I.D., Webb, N.R., and Coulson, S.J. (2002) Primary community assembly on land – the missing stages: why are the heterotrophic organisms always there first? J Ecol 90: 569–577.

Hodkinson, I.D., Coulson, S.J., and Webb, N.R. (2003) Com- munity assembly along proglacial chronosequences in the high Arctic: vegetation and soil development in north-west Svalbard. J Ecol 91: 651–663.

Högberg, P. (1997) 15N natural abundance in soil-plant systems. New Phytol 137: 179–203.

Klingelhofer, G., Morris, R.V., Bernhardt, B., Schroder, C., Rodionov, C.S., de Souza, P.A., Jr, et al. (2004) Jarosite and hematite at Meridiani Planum from Opportunity’s Mössbauer spectrometer. Science 306: 1740–1745.

Klute, A. (1986) Water retention: laboratory methods. In Methods of Soil Analysis, Part 1: Physical and Mineralogi- cal Methods. Klute, A. (ed.). Madison, NJ, USA: American Society of Agronomy, Soil Science Society of America, Publisher, pp. 635–661.

Konhauser, K. (2007) Introduction to Geomicrobiology.

Oxford, UK: Blackwell Science.

Lakanen, E., and Erviö, R.A. (1971) Comparison of eight extractants for the determination of plant available micro- nutrients in soils. Suom Maataloustiet Seuran Julk 123:

232–233.

Lindsay, W.L., and Norwell, W.A. (1969) Equilibrium relation- ship of Zn, Fe, Ca and H with EDTA and DTPA in soils. Soil Sci Am Proc 33: 62–68.

Lipp, J.S., Morono, Y., Inagaki, F., and Hinrichs, K.-U.

(2008) Significant contribution of Archaea to extant biomass in marine subsurface sediments. Nature 454:

991–994.

Lizama, H.M., and Suzuki, I. (1988) Bacterial leaching of a sulfide ore by Thiobacillus ferrooxidans and Thiobacillus thiooxidans. Part I: shake flask studies. Biotech Bioeng 32:

110–116.

López, B.C., Holmgren, M., Sabaté, S., and Gracia, C.A.

(2008) Estimating annual rainfall threshold for establish- ment of tree species in water-limited ecosystems using tree-ring data. J Arid Environ 72: 602–611.

López-Lara, I.M., Sohlenkamp, C., and Geiger, O. (2003) Membrane lipids in plant-associated bacteria: their biosyn- theses and possible functions. Mol Plant Micr Interact 16:

567–579.

Nemergut, D.R., Anderson, S.P., Cleveland, C.C., Martin, A.P., Miller, A.E., Seimon, A., and Schmidt, S.K. (2007) Microbial community succession in an unvegetated, recently deglaciated soil. Microbial Ecol 53: 110–122.

Nichols, B.W., and Wood, R.J.B. (1968) New glycolipid spe-

cific to nitrogen-fixing blue-green algae. Nature 217: 767–

768.

Ogle, K., and Reynolds, J.F. (2004) Plant responses to pre- cipitation in desert ecosystems: integrating functional types, pulses, thresholds, and delays. Oecologia 141:

282–294.

Pansu, M., and Gautheyrou, J. (2006) Handbook of Soil Analysis. New York, NY, USA: Springer.

Puente, M.E., Li, C.Y., and Bashan, Y. (2004a) Microbial populations and activities in the rhizoplane of rock- weathering desert plants. II. Growth promotion of cactus seedlings. Plant Biol 6: 643–650.

Puente, M.E., Li, C.Y., Bashan, Y., and Lebsky, V.K. (2004b) Microbial populations and activities in the rhizoplane of rock-weathering desert plants. I. Growth Root colonization and weathering of igneous rocks. Plant Biol 6: 629–642.

Rensing, S.A., Lang, D., Zimmer, A.D., Terry, A., Salamov, A., Shapiro, H., et al. (2008) The Physcomitrella genome reveals evolutionary insights into the conquest of land by plants. Science 319: 64–69.

Rˇ ezanka, T., Víden, I., Go, J.V., and Dembitzky, V.M. (2003) Polar lipids and fatty acids of three wild cyanobacterial strains of the genus Chroococcidiopsis. Folia Microbiol 48:

781–786.

Santi, C.A., Certini, G., and D’Acqui, L.P. (2006) Direct deter- mination of organic carbon by dry combustion in soils with carbonates. Comm Soil Sci Plant Anal 37: 155–162.

Sasaki, K., and Konno, H. (2000) Morphology of jarosite- group compounds precipitated from biologically and chemi- cally oxidized Fe ions. Can Mineral 38: 45–56.

Schwinning, S., Sala, O.E., Loik, M.E., and Ehleringer, J.R.

(2004) Thresholds, memory, and seasonality: understand- ing pulse dynamics in arid/semi-arid ecosystems. Oecolo- gia 141: 191–193.

Silverman, M.P., and Lundgren, D.G. (1959) Studies on the chemoautotrophic iron bacterium Ferrobacillus ferrooxi- dans. I. An improved medium and a harvesting procedure for securing high cell yields. J Bacteriol 77: 642–647.

Skidmore, M., Anderson, S.P., Sharp, M., Foght, J., and Lanoil, B.D. (2005) Comparison of microbial community compositions of two subglacial environments reveals a possible role for microbes in chemical weathering pro- cesses. Appl Environ Microbiol 71: 6986–6997.

Southam, G., and Beveridge, T.J. (1992) Enumeration of thiobacilli within pH-neutral and acidic mine tailings and their role in the development of secondary mineral soil.

Appl Environ Microbiol 58: 1904–1912.

Sturt, H.F., Summons, R.E., Smith, K., Elvert, M., and Hinrichs, K.-U. (2004) Intact polar membrane lipids in prokaryotes and sediments deciphered by high- performance liquid chromatography/electrospray ionization multistage mass spectrometry-new biomarkers for bio- geochemistry and microbial ecology. Rapid Commun Mass Spectrom 18: 617–628.

Sundareshwar, P.V., Morris, J.T., Koepfler, E.K., and Forn- walt, B. (2003) Phosphorous limitation of coastal ecosys- tem processes. Science 299: 563–565.

Talbot, H.M., Summons, R.E., Jahnke, L.L., Cockell, C.S., Rohmer, M., and Farrimond, P. (2008) Cyanobacterial bac- teriohopanepolyol signatures from cultures and natural environmental settings. Org Geochem 39: 232–263.

Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., and Higgins, D.G. (1997) The ClustalX windows interface:

flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res 24: 4876–4882.

Uroz, S., Calvaruso, C., Turpault, M.P., Pierrat, J.C., Mustin, C., and Frey-Klett, P. (2007) Effects of the mycorrhizo- sphere on the genotypic and metabolic diversity of the bacterial communities involved in mineral weathering in a forest soil. Appl Environ Microbiol 73: 3019–3027.

Van der Wielen, P.W.J.J., Bolhuis, H., Borin, S., Daffonchio, D., Corselli, C., Giuliano, L., et al. (2005) The enigma of prokaryotic life in deep hypersaline anoxic basins. Science 307: 121–123.

Wang, T., and Peverly, J.H. (1998) Screening a selective chelator pair for simultaneous determination of iron (II) and iron (III). Soil Sci Soc Am J 62: 611–617.

Yoshitake, S., Uchida, M., Koizumi, H., and Nakatsubo, T.

(2007) Carbon and nitrogen limitation of soil microbial res- piration in a high Arctic successional glacier foreland near Ny-Ålesund, Svalbard. Polar Res 26: 22–30.

Zielke, M., Ekker, A.S., Olsen, R.A., Spjelkavik, S., and Solheim, B. (2002) The influence of abiotic factors on bio- logical nitrogen fixation in different types of vegetation in the high arctic, Svalbard. Arctic Antarctic Alpine Res 34:

293–299.

Supporting information

Additional Supporting Information may be found in the online version of this article:

Fig. S1. The ML-RC1 site in the moraine of the Midtre Lovénbreen glacier, Svalbard. This site is bigger than site ML-RS1 with 10-m-long rusty leaching strip departing from a rock amphitheater. The soil crust was sampled about every 2.5 m from the origin close to the rock amphitheater to the end of the rusty colour in the grey moraine (not visible).

Acidithiobacillus ferrooxidans most probable number counts, determined on the samples, showed decreasing values from the origin to the end (respectively, 2.3¥ 104, 2.3 ¥ 103, 6.8¥ 102, 0.8 ¥ 101 and 2.3 ¥ 100 cfu g^-1wet soil).

Fig. S2. DRIFT spectra indicating the presence of the mineral jarosite. Relevant peaks used as signature for jarosite are indicated. a, DRIFT spectrum of the soil crust from the rusty strip of site ML-RS1. Several signature peaks of jarosite are indicated. b, DRIFT spectrum of jarosite formed by Acidithio- bacillus ferrooxidans SB1 isolated from site ML-RS1, while growing on 9K medium containing ferrous sulfate.

Fig. S3. Bacterial community structure and diversity in the soil crusts of site ML-RS1. Detrended Correspondence Analysis (DCA) of the ARISA profiles obtained from all the samples withdrawn in four (A to D) minitransects of the site.

Samples were labelled with different colours according to the pH of the soil crust. Circles indicate the soil crust samples, while triangles indicate the soil withdrawn 5 cm below the soil crusts. Letters and numbers accompanying each sample, respectively, indicate the transect and the sample within the transect. Soil crust samples showed a grouping trend accord- ing to the pH, while deeper samples did not. The inset shows the correlation curve between the ARISA-based DCA axis 1 diversity index and the pH of the samples.

Fig. S4. Absolute (mg lipid/g sediment) and relative (%) lipid concentrations from transects A (A) and B (B). Sterols were not analysed in transect B. Abbreviations: heterocyst glycolipids (HG), glycosyldiacylglycerol (Gly-DAG), diglyco- syldiacylglycerol (2Gly-DAG), sulfoquinovosyldiacylglycerol (SQ-DAG), glucuronic acid diacylglycerol (GlyA-DAG), phos- phatidylglycerol (PG), phosphatidic acid (PA), phosphati- dylserine (PS), phosphatidylinositol (PI), phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidyl-(N)- methylethanolamine (PME), phosphatidyl-(N,N)-dimethyle- thanolamine (PDME), ornithine lipids (OL) betaine lipids (BL).

Table S1. Fatty acid composition of IPL-glycerol derivatives.

Major fatty acids found of IPL-glycerides along the two transects A and B as identified with HPLC-ESI-MS in positive and negative ion mode.

Table S2. Chemical, physical and microbiological param- etersa in the greyish, green and rusty soil crusts after removal of plant cover.

Table S3. Diversity indices of bacterial communities in soil crusts of site ML-RS1. Shannon–Wiener and Equitability Indices deduced from ARISA profiles and 16S rRNA gene libraries of the bacterial community in different transects within the site ML-RS1. Means⫾ SD are reported.

Table S4. Results of the screening of 16S rRNA libraries from rusty soil crust (RUS), green soil crust (GRE) and greyish soil crust (GRY). The percentage of identified clones at phylum, class, order, and/or family and/or genus has been reported for each taxonomical rank as the average of samples of transects A and B.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.